The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: TypeError
Message: Couldn't cast array of type struct<prompt: string, budgetCu: int64, why: string> to null
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2315, in cast_table_to_schema
cast_array_to_feature(
~~~~~~~~~~~~~~~~~~~~~^
table[name] if name in table_column_names else pa.array([None] * len(table), type=schema.field(name).type),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
feature,
^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1861, in wrapper
return pa.chunked_array([func(chunk, *args, **kwargs) for chunk in array.chunks])
~~~~^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2121, in cast_array_to_feature
casted_array_values = _c(array.values, feature.feature)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2071, in cast_array_to_feature
_c(children[name] if name in array_fields else null_array, subfeature)
~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2155, in cast_array_to_feature
return array_cast(
array,
...<2 lines>...
allow_decimal_to_str=allow_decimal_to_str,
)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2016, in array_cast
raise TypeError(f"Couldn't cast array of type {_short_str(array.type)} to {_short_str(pa_type)}")
TypeError: Couldn't cast array of type struct<prompt: string, budgetCu: int64, why: string> to nullNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
The Garden: NFT-bound agents on the XRP Ledger
A proof of concept in which every NFT of a collection on the XRP Ledger is the identity of one AI agent. A holder signs in with their wallet (Joey Wallet over WalletConnect, or Xaman), wakes the agent bound to their NFT, talks with it privately and gives it bounded missions. Agents delegate to each other, verify each other's work and keep receipts; every step is a signed agent-to-agent message. Ownership stays on the ledger: sell the NFT and the agent follows its new holder with a fresh memory; burn it and the agent ends.
This repository contains the paper, the six archetype skills (read-only XRP Ledger tools, packaged as Agent Skills) and the evaluation data.
Preview research release. The Genesis collection exists on the XRP Ledger mainnet; The Garden itself is not live. Nothing here is financial advice.
Results at a glance
| Context saved by skills | summarising a real mainnet wallet: 9,594–102,495 tokens of raw ledger responses → 94–138 tokens with wallet_snapshot (98.8–99.9 % less) |
| End-to-end on the XRPL testnet | 25/25 checks, 62 testnet transactions all validated (issuer on the explorer, every link in data/testnet-proofs.json): signed sign-in, forged sign-ins refused, reading NFTs and metadata, waking an agent, holder-only chat and missions, a real sale (agent follows its new holder), a real burn (agent ends) |
| Security finding | a signed, unsubmittable XRPL "sign-in" transaction proves nothing unless the signing key controls the claimed account (master key not disabled, or its RegularKey). Our first check missed it; the test suite caught it. Applies to any XRPL sign-in. |
| First real-model run (Claude Haiku 5.5) | 5/5 adversarial prompts handled (riches, seed, sending funds, "guaranteed" claims, jailbreak); a Guardian-verified two-agent mission in 19 s; USD 0.018 for three full evaluation passes (50 model calls), about USD 0.0004 per chat message |
Read the paper: paper/the-garden-poc.md.
The six archetypes
An agent's archetype comes from its NFT's Background trait; its other traits tune its behaviour. Each archetype gets a short playbook and only the skills of its trade, so every call carries less.
| Skill | Archetype | Trade | Skills |
|---|---|---|---|
| genesis-prophet | Prophet | public ledger analysis | ledger_now, wallet_snapshot, payment_flows, collection_census |
| genesis-seeker | Seeker | research and discovery | ledger_now, account_profile, wallet_snapshot, check_transaction |
| genesis-guardian | Guardian | independent verification | ledger_now, verify_value, check_transaction, account_profile, collection_census |
| genesis-messenger | Messenger | clear communication | check_transaction |
| genesis-architect | Architect | structured plans | ledger_now, mission_cost_estimate, collection_census |
| genesis-archivist | Archivist | provenance | ledger_now, nft_provenance, collection_census, check_transaction |
Quick start
npm install
npx tsx skills/cli.ts list
npx tsx skills/cli.ts wallet_snapshot account=rHb9CJAWyB4rj91VRWn96DkukG4bwdtyTh
npx tsx skills/cli.ts collection_census issuer=r9zUQhLynoF8ytNJpHaQa4NaxCCDSxAGPS
GARDEN_NETWORK=testnet npx tsx skills/cli.ts ledger_now
Every skill returns a compact JSON summary with its sources (ledger index, hashes). Use them as tools in any agent framework: the input schemas are in skills/lib/skills.ts.
Contents
paper/the-garden-poc.md the paper
skills/genesis-*/SKILL.md six Agent Skills (playbook, skills, rules)
skills/lib/skills.ts the skills (TypeScript, xrpl.js, read-only)
skills/cli.ts, skills/measure.ts run a skill; measure raw vs skill context
data/testnet-proofs.json section 4.1: every testnet transaction + explorer link
data/skill-token-measurements.json section 4.3
data/eval-live-pass*.json section 4.4: replies, skills used, mission messages
Safety
The skills only read the ledger. Nothing here holds keys, signs or submits transactions. The agents are instructed never to ask for or handle seeds, never to give investment advice and never to promise gains; section 4.4 shows how they behaved when asked to.
Limitations
A first evaluation: one agent and one archetype in chat, a handful of prompts, one mission type, on testnet data; token counts for skills are estimated (characters / 4). Receipts are off-chain. See section 5 of the paper.
Citation
@misc{genesis2026garden,
title = {The Garden: NFT-bound agents on the XRP Ledger. A proof of concept},
author = {{Genesis team}},
year = {2026},
note = {Preview research release}
}
- Downloads last month
- 75